Many bird radar studies provide estimates of the number of birds flying past a given area, but very few of these actually estimate detectability. One of the challenges in radar ornithology is estimating the probability of detection of flying targets with altitude and distance. I estimated the detection patterns associated with three marine radars by using a combination of field trials and simulation modelling, and estimated the probabilities of correctly and incorrectly detecting birds in relation to altitude. The results indicate considerable variation in power among radar units. The nominal beam width was 4 degrees and effective beam width was 7 degrees. The results from the simulation indicate detectability varies with altitude, with few birds detected in the lower altitude bands. Many simulated birds were classified as two different birds when crossing the beam twice and there were many false detections, especially in the lowest altitude bands near the radar.